American Mineralogist, Volume 69, pages 69-78, 1984

Fugacities of sulfurous gases in pyrrhotite-bearing silicic magmas

Jeur,s A. WHrrNev

Department of GeologyUniversity of Georgia, Athens, Georgia 30602

Abstract

Thermodynamic data are used to determine the fugacities of sulfurous gases 52, SO2 andH2S along the pynhotite saturation surface within the system Fe-S2-O1SiO2. Thefugacities of these species are contoured on diagrams of log /o2 versus T ('C). Thesediagrams allow the fugacities of sulfurous species for silicic magmas to be estimatedassuming that the temperature and oxygen fugacity are determined from coexistingmagnetite and ilmenite, and the pyrrhotite inclusions represent magmatic conditions. Incases where the pyrrhotite was analyzed and the fugacities calculated, the results areconsistent. This method is widely applicable because it may be used for any rock for whichiron-titanium oxide data are available and igneous pyrrhotite is present.

Both 52 and SO2 are highly dependent on /o2 and T, decreasing with both variables.Sulfur ranges from l0 to l0-a bars whereas SO2 varies from ld to 10-3. Hydrogen sulfide isless variable, ranging between 100 and l0 bars, but it also decreases with log fo2 andtemperature.

The initial volatile phases evolved at high temperatures from oxidized magmas are rich inSO2. Volatile separation depletes the magma in sulfur and causes reduction of the magmaby oxidation of HS- in the melt to SO2 in the vapor phase. Such magmatic volatile phasescan be an important source of sulfur for some types of ore deposits.

Introduction

Sulfur dioxide and hydrogen sulfide have long beenrecognized as important, and odoriferous, components ofvolcanic emanations. Until recently, the activities ofthese species have not been quantified in magmas respon-sible for major silicic volcanism or plutonism, althoughthe high SO2 activities implied by certain assemblageswere noted by several authors (Popp el al.,1977;Tsu etal., 1979). The recognition of pyrrhotite as inclusions inphenocrysts of ash-flow tuffs (Hildreth, 1977; Drexler,l9E2; Whitney and Stormer, 1983) allows the activities ofthese species to be calculated for some occurrences andgives us insight into the importance of sulfur in calc-alkaline silicic magmas prior to near-surface degassing.

When the composition of the pyrrhotite can be careful-ly determined, the sulfur fugacity can be calculated(Toulmin and Barton, 1964). This value, combined withthe fugacity ofoxygen and water, and temperature deter-mined from other mineral relationships, allows the fugaci-ties of SO2, SO:, and H2S to be calculated. Unfortunate-ly, the composition of the pyrrhotite is difficult todetermine with sufficient accuracy (+0.1 moleTo FeS)because of the small size of the inclusions, spontaneousexsolution on quenching, and the effects of oxidationduring cooling and devitrification. Therefore, the direct

0003-004)084/0102-0069$02.00 69

determination of sulfur fugacity can only be done on thefreshest, least oxidized samples with large inclusions.

The position of the pyrrhotite-magnetite boundarywithin the system Fe-OrSz-SiO2 gives an alternativemethod of calculating the sulfur fugacity if the tempera-ture and oxygen fugacity are known, as is the case whenco-existing iron-titanium oxides are present (Buddingtonand Lindsley, 1964; Spencer and Lindsley, l98l). Thispaper develops diagrams from which the fugacity ofsulfurous species may be determined assuming primarypyrrhotite inclusions are present, an iron silicate ofknown thermodynamic character is present, and thetemperature and oxygen fugacity are known.

Pyrrhotite occurrence

Pyrrhotite in silicic volcanics is found as round inclu-sions within phenocryst phases (Hildreth, 1977;Drexler,1982; Whitney and Stormer, 1983). It has been observedas inclusions within magnetite, mafic silicates, sphene,and plagioclase. It rarely forms euhedral crystals, withthe dominant morphology being rounded blebs. To date,it has not been observed or reported within glass. Figure Ishows typical pyrrhotite inclusions in magnetite. Becausethe round blebs are found in iron-free plagioclase as wellas mafic minerals, the shape is an original morphology

70 WHITNEY: SULFUROUS GASES IN PYRRHOTITE SIUAC MAGMAS

Fig. l. Typical pyrrhotite inclusions within magnetite fromthe Fish Canyon Tuff, Central San Juan volcanic field, SouthernColorado (Whitney and Stormer, 1983).

rather than the result of reaction with the host. This shapemay be caused by resorption by the melt before inclusion,or it may be an original growth morphology or pseudo-morphs after blebs of an immiscible liquid (Whitney andStormer, 19E3).

The best preserved crystals are found in vitric volcanicrocks. During cooling and devitrification the pyrrhotiteoxidizes to pyrite and magnetite by reaction with theatmosphere. An idealized end-member reaction is:

6FeS + 2O2 :3 FeS2 * Fe rO+ . ( l )

Such oxidation can cause an excess of sulfur in theremaining sulfide and destroy the original composition.Sulfur rich pyrrhotites with less than 0.47 iron atoms performula cannot even be quenched in the laboratory, butspontaneously exsolve to a mixture of lower temperaturepyrrhotite and pyrite (Corlett, 1968; Barton, 1969; Craigand Scott, 1974). Therefore, what probably exists in mostcases is a fine-grained intergrowth of two phases. Withinmagnetite, reaction with the host occurs quickly as mag-netite begins to re-equilibrate with the exsolution ofulvcispinel or ilmenite accompanying oxidation. Pyrrho-tite enclosed within silicate minerals survives longer andcan give reliable results in devitrified samples, but greatcare must be used to assure that the composition ana-lyzed is representative of the original. In many samples,reliable compositions cannot be obtained.

The small size of the inclusions (often less than l0 r.cm)makes microprobe analysis difficult, especially when theyare in iron-bearing phases. Often, only the largest inclu-sions can be analyzed.

Assuming accurate analyses can be obtained, the sulfurfugacity can be calculated from the composition of thepyrrhotite (Toulmin and Barton, 1964). The results ofthese calculations are sensitive to the ratio of iron tosulfur, and thus the analytical uncertainty in microprobeanalyses (approximately *2Vo) causes more than an orderof magnitude error in suHur fugacity. This uncertainty

leads to larger errors in SOz, and significant uncertaintyin H2S.

Due to all these problems, it is desirable to develop analternative method for use when the original pyrrhotitecomposition is altered or accurate analyses are not possi-ble. The position of the magnetite-pyrrhotite boundarywithin the system Fe-OrSrSiO2 provides such a meth-od. In volcanic rocks, the temperature and oxygen fugac-ity are often known as well as any intensive magmaticvariables due to the occurrence of coexisting magnetiteand ilmenite solid solutions (Buddington and Lindsley,1969; Spencer and Lindsley, l98l). These variables,combined with the position of the pyrrhotite-magnetiteboundary in the simple system, can be used to estimatethe fugacity of various sulfurous species without relyingon the pyrrhotite analyses. Such a method also allows acheck on the results of pyrrhotite analyses.

Phase diagrams for the system Fe-O2-SrSiO2

The most useful diagram for the present purpose is thatin which the fugacities of 02 and 52 are used as coordi-nates. The details of constructing such diagrams havebeen discussed by many authors (e.9., Garrels andChrist, 1965; Holland,1959,1965). Sources of thermody-namic data are summarized in Table l. Isothermal dia-grams for vdrious temperatures are shown in Figure 2.

The calculation of the pyrrhotite-magnetite curve is ofspecial importance for current application. Because pyr-rhotite varies in composition with changing sulfur fugac-ity (Toulmin and Barton, 1964), this boundary is slightlyconcave to the lower left. It originates at a triple point(magnetite-triolite-iron), which is metastable above570'C due to the stability of wiistite and metastable at alltemperatures in the presence of quartz. For our purposes,however, this triple point may be used as the origin of thecurve throughout the temperature range examined be-cause the thermodynamic properties of all three phasesare well known. The pyrrhotite-magnetite boundary maybe constructed graphically by determining the slope of theline for the composition of pyrrhotite stable at a certain

Table l. Sources of thermodvnamic data

Equlllbrla Source of data

2 F e + 5 2 = 2 F e S

3/z re + oz = !/z Fef4

4 F e 3 0 4 + 0 2 = 6 F e 2 Q 1

sz = 2 s(r)3 Fe3OO + S, = FeS, + 4 F62O32/5 FeJOU + sz = F€s2 + 2/5 tdfi4

F" l -*s * T t " - (1-x) Fesz

I /2 s2+ oz- szt/z s" + Ha = Hrstt" + t/z 02 - Hzo

2 Fe)oU+ 3 s102 = 3 Fe2slo4 + 02

1/2 s, + )/2 02 = s}a

Barton ild Sklmer, 1979

Barton and Sklnner, 1979

Berton and Sklmer, 1979

Barton ud SklnnEr, 1979

Barton and Sklnner, 19?9

Barton and sklmer, 1979

T@lnln ud Barton, 1961+

Roble g! aL,, 79?9

Robte g! aJ,, 79?9

Roble gL aJ,, t9?9

Roble g! d,.t 79?9

Roble gq e., L979

WHITNEY: SULFUROUS GASES IN PYRRHOTITE SILICrc MAGMAS 7 l

Fig. 2. Isothermal, quartz saturated, log fo2 vs. log 6zdiagrams for the system Fe-OrSrSiO2. Calculated as describedin the text and Appendix from sources listed in Table l. Thesulfide saturation surface is shown in gray. S(L; = ru11u. ttqu'd,Po.. = pyrrhotite solid solution, Fay : lutultte, Mag =magnetite, Hem = hematite, Py = pyrite.

sulfur fugacity, drawing a short section ofthe curve as astraight line, changing the composition of the pyrrhotite,and repeating the process. Alternatively, the curve maybe calculated using equation (8) of Toulmin and Barton(1964) and the coordinates ofthe triple point (see Appen-dix). Both methods were used and gave satisfactoryresults.

These diagrams may be further contoured for fugacityof SO2 and HzS (Fig. 3). To determine H2S some measureof hydrogen fugacity must be made. It is convenient toestimate the fugacity of H2O and use its dissociationconstant to give -fHz as a function of /o2 and T. As aconsequence, the H2S isopleths vary with water fugacity.The intersections of the contour lines with the sulfideboundary on these diagrams may be transferred to an fo2and T diagram to represent the values along the sulfidesaturation surface,

Figure 4 summarizes the various reactions encounteredin the Fe-O2-S2-SiO2 system projected onto a logfoT-T

diagram. Also shown are the approximate /orT condi-tions for various pyrrhotite-bearing volcanic rocks. Mostof these magmas are moderately oxidizing, with theexception of the Bishop Tutr AII fall within the calculatedstability field of pyrrhotite, which is consistent with theobserved paragenesis. Several occurrences show a morereducing trend at lower temperatures than would beexpected from simple movement along magmatic bufercuryes that are controlled by fixed ferric/ferrous ratios inthe melt (Sack er a/., 1980). Carbon species oxidation-reduction lie at far more reducing conditions than theserocks indicate, and thus methane, graphite, or carbonmonoxide play trivial roles in these silicate melts.

Figure 5 shows the fugacity of sulfur contoured on thesame log/o2-I diagram. Thefz isopleths are flatter thanthe simple buffer curves of Figure 4, and are probablyflatter than the more complex buffers that control silicatemagma crystallization. Therefore, lowering temperatureslead to lowering sulfur fugacities. The approximate range

Fig. 3. Isothermal log/oz versus log$2 diagrams contouredfor logfio2 and log/uzs. Stability fields are the same as in Fig. 2,with contours calculated using data from Robie et al. (1979).Fugacity of HzO is 2fiD bars.

72 WHITNEY: SULFUROUS GASES IN PYRRHOTITE SILIAC MAGMAS

octlo

J

800rec)

800TPC)

Fig. 4. Log/o2-Idiagram for the sulfide-saturated equilibria.Reactions involving FeSOa are only approximately known andare labeled as follows: A. FeSOa* S(L) = pyrite; B. FeSOo =magnetite + pyrite; C. FeSOa = magnetite + S(L). Conditionsfor various pyrrhotite-bearing volcanic rocks are shown. B =Bishop Tutr, Hildreth (1977, 1979); FC : Fish Canyon Tuf,Whitney and Stormer (1983); J = Julcani vitrophyres, Drexler(1982); SH = St. Helens ash, Melson and Hopson (1981). All ofthe above were recalculated from coexisting iron-titaniumoxides using the equations of Spencer and Lindsley (1981) andStormer (1983). Also shown is the approximate conditions of theEl Chich6n ash. Although ilmenite is absent, the presence ofsphene in place of ilmenite restricts the possible oxygen fugacity.Phenocryst assemblage and approximate conditions taken fromLuhr et al. (1982).

ofvalues is l0 to l0-2 bars at 950"C down to l0-r to l0-4bars at 700'C.

Figure 6 shows a similar projection of SO2 contours. Inthis case, the curves are even more closely spaced so thata small decrease in oxygen fugacity makes a drasticdecrease in/so2. A change of0.5 log/o2 changes log/so2by 1.0 at constant log/sz. The fugacity of SO2 tends todecrease with decreasing temperature along standardmagmatic buffer curves, with an average range of valuesdecreasing from 103 to l0-l at 950"C to l0 to l0-3 at700"c.

The fugacity of H2S (Fig. 7) is much less variable,always between l0 and 100 bars, and is dependent onwater fugacity. The diagram shown here is for 2fi)0 bars/u2o. Values may be corrected to other water fugacitiesby adding the term log(fu2o12000) to the raw values. The/n2s again tends to decrease with decreasing temperaturealong magmatic bufer curves, but the change is muchsmaller in magnitude.

Fig. 5. Log/o2vs. T diagram portraying contours of constantlog/s2 on the sulfide saturation surface. Reaction lines are thesame as in Fig. 4.

Perturbations caused by pressure and solid solution

Pressure conection

The pyrrhotite-magnetite curve is shifted slightly bychanges in confining pressure. This etrect is a result oftwo factors: the change in the origin of the curve con-

T(.C)

Fig. 6. Loglo2 vs. Idiagram portraying contours ofconstantlog$o2 on the sulfide saturation surface. Reaction lines are thesame as in Fig. 4.

. t /

Sulfide SolurolionSurtoce

WHITNEY: SULFUROUS GASES IN PYRRHOTITE SILICIC MAGMAS

Fig. 7. Log/o2 vs. Idiagram portraying contourc ofconstantlog/uzs on the sulfide saturation surface for a water fugacity of2000 bars. To convert to other water fugacities add the term logfn2sl2ffi. Reaction lines are the same as in Fig. 4.

trolled by the magnetite-iron and troilite-iron equilibria,and the change in pyrrhotite composition for a specificsulfur fugacity resulting in a change in the slope of thecurve. The first effect is by far the most significant, and isthe only one causing significant perturbations at crustalpressures (see Appendix for detailed derivation of thecurve).

The equilibria controlling the starting point of the curvecan be corrected for confining pressure by adding a PAVterm to the Gibbs free energy of the reaction. These termsfor both equilibria are listed in Table A-2 (Appendix) as afunction of temperature. The shift in the boundary itselfin terms of the change in logf2 per kilobar at constant log/o2 is tabulated in the fourth column of Table A-2 (seeappendix for derivation and tabulation). This value will berepresented in subsequent equations as the parameter(A).

To correct the values of logf2 obtained from Figure 5for pressure, a term must be added that has approximate-ly the magnitude A . P where P is confining pressure inkilobars. Both log/so2 and log/n2s must be corrected byadding a term Al2. P (see Appendix for derivation).

Solid solution in magnetite and pyrrhotite

Solid solution in magnetite, dominantly in the form ofulv<ispinel, shifts the magnetite-pyrrhotite boundary tohigher sulfur fugacities by decreasing the activity ofmagnetite. Similarly, solid solution of other cations foriron in pyrrhotite shifts the boundary to slightly lower

sulfur fugacities (less than 0.05 log/s2 for natural pyrrho-tites). These changes can also be modelled by consideringthe effects of such solid solutions on the magnetite-ironand troilite-iron starting point for the curve (see Appen-dix). If these shifts in the curye are given as changes in log

.1t2 for a fixed log fo2, they take on the following values(see Appendix for derivation).

A(logJtz)sorio solurion = 2log aro - lro,

o^* (2)

To a good first approximation, the o^*irurbe replacedby the mole fraction of magnetite in the magnetite solidsolution calculated by an appropriate method (see forexample Stormer, 1983). The deficiency of iron in thepyrrhotite is already accounted for in the algebraicexpression for the curve (see Appendix). Therefore, theonly parameter affecting the activity of pyrrhotite is thesubstitution for iron. In addition, the apo may be replacedby the number of moles of iron divided by the totalnumber of cations. The resulting correction takes theform:

A(log.Jt2)5o1;6 solution : 2 log XF3 -

where

XF8 :Nr"

N Cations

The corrections for both log"fio2 and log/H2s are l/2 thevalue for log fs2 (see Appendix for detailed derivation).

Procedure for calculating fugacities

In order to obtain the most accurate estimates offugacity for 52, SOz, and H2S from the figures describedin the previous section, the following procedure should befollowed:

l. Carefully examine polished sections to insure thatpyrrhotite appears to be a primary phase in equilibriumwith the phenocryst assemblage.

2. Determine the composition of coexisting magnetiteand ilmenite solid solutions using standard microprobetechniques or other analytical methods.

3. Calculate the temperature and oxygen fugacity ofequilibrium using the most recent solution model formagnetite-ilmenite solid solutions and an appropriatemethod for calculating mole fractions (currently Spencerand Lindsley, l98l and Stormer, 1983 are recommended).

4. Plot the coordinates derived on Figures 5, 6, and 7 toobtain raw values of logfz, log73o2, and logfi2s. Thesevalues will be referred to as (log/sJ', (log/soz)', and (log

-frzs)'.5. If possible, estimate the probable water fugacity and

total confining pressure on the magma during phenocrystequilibration. In many cases only crude approximations

ito, "*&. (3)

(4)

74 WHITNEY: SULFUROUS GASES IN PYRRHOTITE SILIAC MAGMAS

are available, but because the perturbations are small theresults are still satisfactory.

6. If possible, determine the approximate compositionof the pyrrhotite to check for the presence of significantcopper or nickel substituting for iron.

7. Correct the raw values according to the followingequation:

log"fsz : (log"fsJ' + A . P(kbar)

t

- ttotxpsbt 2 loe xF! (5)

log/so2 = (og/so2)' + 4' r(luar)

I

f log XFS,sol log XP.3 (6)

log/H2s = (log/n2s)' + rosffi + {. r(tu*)

I

: log xFS,gof log XF8. (7)

The values ofA for these equations should be selected forthe appropriate temperature from Table A-2. Other pa-rameters are as defined in previous sections.

Calculations for specific occurrences

Table 2 summarizes the results for a number of silicicrocks for which pyrrhotite was reported as a primaryphase. Values in parenthesis were calculated from thepyrrhotite analyses and are approximate values due to theuncertainties previously described. In each case, thesevalues vary by an order of magnitude depending on theactual analysis used. The correlation between the twomethods is therefore good within the uncertainty inherentin both methods. The corrections for pressure, solidsolution, and water fugacity are small, generally less than0.2 log units. Therefore, even in cases where these

variables are poorly known, the raw values for fugacitiesare still useful estimates.

The uncertainties in both methods are significant andshould be considered in comparing data from differentsources. When calculating the sulfur fugacity directlyfrom the pyrrhotite composition, it must be assumed thatthe analysis represents the original bulk composition andno exsolution or oxidation has changed the bulk composi-tion significantly. Such occurrences are rare, but assum-ing this is the case the normal uncertainty in microprobeanalyses is still t2Vo of the amount present. This error inpyrrhotite composition translates into approximately a+1.5 uncertainty in log"fsz at moderately high oxygenfugacity. The calculation of log/so2 and log/u2s are lesscertain because the values of /o2 and /Hzo must beestimated from other mineralogical relationships.

Using the method described in this paper there are twoimportant sources of error. The most important is theuncertainty in oxygen fugacity and temperature inherentin the magnetite-ilmenite geothermometer. According toSpencer and Lindsley (1981) this error is 40 to 80" and 0.5to 1.0 log/o2. These, however, are not independent asanalytic errors cause movement of both variables. Theirestimated uncertainty, when plotted on Figure 5, yieldsan uncertainty of less than tl.0 in log/s2. A secondsource of error is in the position of the pyrrhotite-magnetite curve. During the course of this study, variousthermodynamic parameters were used to calculate thecurve. In no case was the discrepancy more than -r0.2 inlog "fsz. The combined uncertainty using the magnetite-pyrrhotite boundary therefore appears to be equal to, orless than, the analytic uncertainty inherent in the analysisof pyrrhotite. More importantly, the current method isapplicable to cases where the pyrrhotite has exsolved oroxidized, and may be applied in any case where primarypyrrhotite is present and fo2 and temperature can beestimated.

Using the diagrams of this study, maximum limits caneven be placed on the fugacity ofthese sulfurous gases in

Table 2. Fugacity of sulfurous species in parent magmas for some pyrrhotite-bearing silicic volcanic rocks.

r(oc) Log re, ml#,"8-, r;l#:'?k: c#:+= rs2 fsoz fltzs

Bi-shop

Flsh Canyon

Julcanl

St. HeIenE

El Chlchdn

1A

(4o)75

(80)9+(.rs)),t

4?

7m

800

880

950

850

-L4,4

-77,6

- 9 . 8

-L0,2

- 1 1 . 0

-2 ,9 - 2 .7 - r , 6 - L ,5 +L .47 +L .62 o .oo2 O .O3(o .oo5 ) ( 0 .10 )

-0 ,2 +o .2 +2 ,L +2 ,2 +1 ,59 +1 .88 f . 6 76(z.o) (rto)

+0 ,8 +0 .9 +3 ,2 +3 ,2 +1 .78 +7 .?3 8 1600G) (.rso)

- 0 ,6 - 0 .4 +L ,2 +L , ) +7 ,77 + ! , 52 0 .4 20

-O ,2 0 .0 +2 .0 +2 ,L +7 ,77 +7 ,67 1 ,0 L25Noter A11 fugacltles are 1n bars. Nunbers 1n parentheses are average values calculated frcn chenical analyses of

Wrhotite. All other estlmtes iletemlned as alescrlbed in thls lalEr. Data recaLculated frcn the folloillngsourcesr Blsho!, Hllalret|n (Lgn)t Fish Canyon, lthltney and Stomer (1983)t Julcanl, Drexler (1982)t St. Helens,Melson ard Hopson (1981); El Chtch6n, Luhr et al. (fSA2). l l f tenlnmtures sd log fg2.values recaiculated. fronthe orlglnal alata uslng the expresslon of Spencer and Lindsley (1981) ad Stomer (1p82), tr\rgaclty of watereetlmted 1n cllfferent ways by vulous authorc. E1 Chlch6n data is only apprcxlnate due to the lek of llnenlte 1nthe phenocryst assenblage.

WHITNEY: SULFUROUS GASES IN PYRRHOTITE SILIAC MAGMAS 75

volcanic rocks that do not have primary pyrrhotite. Theabsence means their fugacity falls below the boundary. Ittherefore appears that the current method is at least asaccurate as calculations based on microprobe analyses ofpyrrhotite, and is much more widely applicable.

Role of sulfurous gases in silicic magmas

Examination of Figures 5, 6, and 7 demonstrates thatSO2 and H2S are the dominant sulfurous species in gasesequilibrated with pyrrhotite-bearing magmas. Calcula-tions of SO3 fugacity show it to be from 2to l0 orders ofmagnitude less than SO2, so that it is never a dominantspecies. The values obtained for SO2 may be quite highfor oxidizing magmas, and in some cases it may be adominant component of the gaseous phase. Hydrogensulfide is far less variable, generally being in the tens ofbars. It is significant for all conditions of/o2 and T.

Under closed system conditions, the fugacities of allsulfur species decrease with decreasing temperaturealong normal magmatic buffer curves through the precipi-tation of pyrrhotite. This pattern permits pyrrhotite toremain stable; however, separation of the volatile phaseseverely alters this process.

As a silicic magma rises toward the surface, earlyvolatile phases will have approximately the fugacitiescalculated for the parent magma. The magma, however,probably has a rather low abundance of sulfur dissolvedin the melt (Burnham, 1979). Therefore as early volatilephases separate, the melt is rapidly depleted in sulfur.Under oxidizing conditions, the amount of SO2 in earlyvolatile emanations may be rather large, but very little ofthis component can be derived from a sulfur-poor melt.Any sulfide in contact with the melt would probably reactto release sulfur. This process may explain why pyrrho-tite is not reported in the glass of silicic volcanic rocks,but is only included in other phases. Certainly, themajority of the sulfur dissolved in the melt would bereleased during degassing accompanying eruption, so thatwhole-rock sulfur values of tuffs are much lower than thevalues in the parent magma.

Although there is no direct evidence as to the sulfurspecies present in the melt, Burnham (1979) has hypothe-sized that the dominant one is HS- based on H2O andCO2 as analogs for H2S and SO2 respectively. This as-sumption awaits experimental confirmation, but if it istrue, separation of an SO2-rich gas phase would takeoxygen from the melt to form SO2, causing a reduction info2 even at constant temperature. This phenomenon mayexplain why analyses from Julcani and the Fish CanyonTuff (Fig. 4) show a trend to lower oxygen fugacities atlow temperatures greater than would be expected forcommon magmatic buffer curves.

Early-formed magmatic volatiles rich in SO2 may beimportant sources of sulfur for metallic ore deposits.These volatile phases would also preferentially concen-trate chlorine from the magma, which would help trans-

port metallic elements in solution. Oxidizing magmaticconditions evidenced by the early formation of sphenewould be beneficial to the formation of such solutions.

One important question is the source of sulfur for thesesilicic magmas. In the exceptional case of El Chich6n, thehigh SO2 content may have been derived by assimilationof sulphates from evaporites in the upper crust (Luhr etal., 1982). In other cases, sulfides deposited in volcaniccrust by hydrothermal circulation at ocean ridges may bean important source of sulfur. If the examples studied sofar are typical, then many calc-alklaine magmas weresaturated with pyrrhotite at an early stage and must havehad a sulfide-rich protolith. Altered oceanic crust alsohelps explain the oxidized conditions of these magmas, asoxidation accompanies both the reduction of sulphate insea water in hydrothermal systems, and the conversion ofpyrite to pyrrhotite during metamorphism (Whitney andStormer, 1983).

Conclusions

The fugacity of sulfurous gases in pyrrhotite-bearingmagmas can be estimated from temperature and oxygenfugacity with an accuracy equal to, or better-than, calcu-Iations based on analyses of the pyrrhotite composition.In addition, the method outlined herein is not affected bythe small size of the pyrrhotite or exsolution and oxida-tion accompanying eruption and deposition.

The sulfur fugacity in pyrrhotite-bearing magmas isbetween l0 and 10-a bars and decreases with temperatureand oxygen fugacity. The fugacity of sulfur dioxide isbetween 103 and l0-3 bars, also decreasing with tempera-ture and oxygen fugacity. Hydrogen sulfide fugacity isless variable, between l0 and 100 bars, but is dependenton water fugacity.

The initial volatile phase evolved at high temperaturesfrom oxidized magmas is rich in SO2, Separation of thevolatile phase rapidly depletes the melt of sulfur, greatlylowering the whole-rock sulfur content and may reducethe oxygen fugacity in the melt below that expected forbuffer curves controlled by constant ferric-ferrous ironratios.

All volcanic materials should be carefully studied toidentify the occurrence of pyrrhotite as it appears to berather common in a number of calc-alkaline silicic rocks.Its occurrence has important consequences for the originand evolution of such magmas.

Acknowledgments

This research was supported by NSF grant no. EAR-8204568and with the support and cooperation of the U.S. GeologicalSurvey. Part of the mathematical derivation was inspired by T.Stanley. Helpful ideas and contributions were made by J. C.Stormer, Jr., W. Rose, and J. W. Drexler. The manuscript wasgreatly improved by reviews from P. Barton, J. Craig, and M.Cameron.

76 WHITNEY: SULFUROUS GASES IN PYRRHOTITE SIUCrc MAGMAS

ReferencesBarton, P. B., Jr. (1969) Thermochemical study of rhe system

Fe-As-S. Geochemica et Cosmochimica Acta, 33, 841-857.Barton, P. B., Jr. (1970) Sulfide petrology. Mineralogical Society

of America Special Paper Number 3, 187-198.Barton, P. B., Jr. and Skinner, B. J. (1979) Sulfide mineral

stabilities. In Barnes, H. L., Ed., Geochemistry of Hydrother-mal Ore Deposits, 2nd Editions. p. 278403, Wiley, NewYork.

Buddington, A. F. and Lindsley, D. H. (t964) Iron ritaniumoxide minerals and synthetic equivalents. Journal of Petrolo-gy, 5, 310-357.

Burnham, C. W. (1979) Magmas and hydrothermal fluids. InBarnes, H. L., Ed., Geochemistry of Hydrothermal OreDeposits, 2nd Edition, p. 7l-136, Wiley, New york.

Corlett, M. (l%8) Low-iron polymorphs in the pyrrhotite group.Zeitschrift fiir Kristallographie, 126, 124-134.

Craig, J. R. and Scott, S. D. (1974) Sulfide phase equilibria. InMineralogical Society of America Short Course Notes, Vo!ume l, Sulfide Mineralogy, p. CS-l to CS-110. WashingronD.C.

Drexler, J. W. (19E2) Magmatic conditions from vitric units ofthe Julcani district, Peru. PhD. dissertation (unpublished),Michigan Technological University, Houghton, Michigan.

Garrels, R. M. and Christ, C. L. (1965) Solutions, Minerals, andEquilibria. Harper and Row, New York.

Hildreth, E. W. (1977) The magma chamber of the Bishop Tuf:Gradients in temperature, pressure, and composition. PhD.dissertation (unpublished) University of California, Berkeley.

Hildreth, E. W. (1979) The Bishop Tutr. Evidence for the originof compositional zonation in silicic magma chambers. InChapin, C. E., and Elston, W., Eds., Ash Flow Tuffs, p. 43-75, Geological Society ofAmerica Special Paper 180.

Holland, H. D. (1959) Some applications of thermochemical datato problems of ore deposits. I. Stability relations among theoxides, sulfides, sulfates and carbonate of ore and gangueminerals. Economic Geology, 54, 184-231.

Holland, H. D. (1965) Some applications of thermochemical datato problems of ore deposits. II. Mineral assemblages and thecomposition of ore-forming fluids. Economic Geology, 60,I l0l- l 166.

Luhr, J., Carmichael, I. S. E., and Varekamp, J. C. (1982)eruption of El Chich6n volcano, Chiapas Mexico (abstract)EOS (Transactions of the American Geophysical Union), 63,rr26-tr27.

Melson, W. G. and Hopson, C. A. (1981) Pre-eruption tempera-tures and oxygen fugacities in the l9E0 eruption sequence. InLipman, P. W. and Mullinax, D. R., Eds., The 1980 Eruptionof Mount St. Helens, Washington, U. S. Geological SurveyProfessional Paper 1250, p. 641-648.

Popp, R. K., Gilbert, N. C., and Craig, J. R. (1977) Stability ofFe-Mg amphiboles with respect to sulfur fugacity. AmericanMineralogist, 62, 13-30.

Robie, R. A., Hemingway, B. S., and Fisher, J. O. (1979)Thermodynamic properties of minerals and related substancesat 298.15 K (25"C) and one bar (105 pascals) pressure and arhigher temperatures (revised 1979). U.S. Geological SurveyBulletin 1452. (originally printed in 1978).

Sack, R. O., Carmichael, I. S. E., Rivers, M, and Ghiarso, M. S.(1980) Ferric-ferrous equilibria in narural silicate liquids at Ibar. Contributions to Mineralogy and Petrology, 75,369-376.

Spencer, K. J. and Lindsley, D. H. (1981) A solution model for

coexisting iron-titanium oxides. American Mineralogist, 66,I 189-1201.

Stormer, J. C., Jr. (1983) The efects of recalculation on esti-mates of temperature and oxygen fugacity from analyses ofmulticomponent iron-titanium oxides. American Mineralogist,68, 5E6-594.

Toulmin, P., III and Barton, P. B., Jr. (l9U) A thermodynamicstudy of pyrite and pyrrhotite. Geochimica et CosmochimicaActa ,28 , U l -671.

Tsu, J. L., Gilbert, M. C., and Craig, J. R. (1979) Sulfidation ofsynthetic biotite. American Mineralogist, 64, 3M-316.

Whitney, J. A. and Stormer, J. C. Jr., (1983) Igneous sulfides inthe Fish Canyon Tuff and the role of sulfur in calc-alkalinemagmas. Geology, ll, 99-102.

Manuscript received, December 13, 1962;acceptedfor publication, July 19, 1983.

ApIEnaExr Tlrenodyrunlc D€tenlEtlon of tlte

4Ehotlte-Uagnetlte Eounda4r

Th€ lD61t1on of the Dagneilte-pyrhottte b@ndary rlthln i,he

sxsten Fe-S2-0, rEJ b€ dotenlneil algebElcally drulDg thai one

polnt on th€ cuwe 16 knom, ud that the conlEltlon of the ItJrrrhe

tite le knom relatlve to th€ ill-fur fugrclty.

The Factlon of the curye rEy he yrltt€n ln th€ fom

(8) 3 FexS + a< 02 ' x re3o4 + 3/Z S"

ilheF x 16 th€ nol€ fEctlon of FeS ln a fom]a unlt of lDrrrhotlte

Flthln th€ systen Fe-S. the slolE of the rEgn€tli€-pyrhotlt€

boundary on the log fsz-log foz dlagru cs be ilerlved fM the

equlllbrLM constut equatlonr

( 9 ) l o g K = ) / z t o e f S 2 + x l o g . r * - O l o g f O 2 - 3 l o g e p o ,

(10) 2x 1og foz = - 1og K + 3/2 Iog fs} ,

as€unlng that the actlvlty of, nagnetlte ud pyrhotlte e unlty ln

the fomulas as Eltten, Dlfferentlatlng y1e1d6 the slolE of the

cuwe I

/ { , - \ d ( l o g f 9 , )\rra/ .rr.-=i", = # , o" (ub) .d(ros to) =

i* a(ros fsr)cr r rog rs r , / qx '

Integratlng (rro] ,tor sone startlng polni (o) to s@6 enal poLnt (n)

rlLl detemlne the charge ln log fg, along th€ cune!

(1os fsz)n

4l a(ros rsr)\ rog rs2 /o

slnc€ th€ log fg, ls knom d a functlon of nole frutl-on of pyrho-

tlte (N ln eqEtlon I of Toulnin sd Berton, L96+), ftu chaln nle

nal bs uaed to chang€ the Erlabl€ of lntegratlonr

d(ro8 rs.,/(14) a(ros fsr) = -6.- dN

1(ros fo2)n

(12) ) a(ros rs.)I (to6 tsa).

(8) (rog fsr), - (]os foz)o =

WHTTNEY: SULFUROUS GASES IN PYRRHOTITE SILICIC MAGMAS 77

( N n - , -

(1i) (rog rsr), - (1og rsr). = \ d

tt'ff tt" *

/ * o

The log fga Is glven by T@lnln ed Buton (1961+, equatlon 8) asr

(16) 1os rs2 = (rc,oJ - Bs.a) r)(!F - t1 + 393t/To,99at *- - tt,9t,

{here T 1s in oK and N 1s the nole fEctlon of FeS ln pyrhottte

*1th1n the systen FeS-S2, However, slnce Toulnln dd Barbon deflneat

N Htthln the syGten FeS-SZ, rather thaa FeS-S as done tn equatton (8),

the x ln equatlon (1J) nust also be replaced by 1t6 egul%lsnt 1n

tena of N. This equallty tak€s the fom

(ua) I =;;ft;6- , (rzu) ', - A

Dlffeenttatlng cquatlon (16) and aubstltutlng the reaults along wlth

(1?b) lnto (1J), we gst the fol lowlngl

, .o, , a(tog fso) _1^.(16 ) - : : ? : =_s5 .s5 (10$o_ r l - # ;

fnn r(rg) (roe rer)n - (log ror)o = \ ? t#l J-er.431ff - t) - SgLl*

/ N -

Ir*.( zo ) = -e i , . 371+- r l f \ f r a r -

L ' N O

. ' l N n r -(21) = -&t?(Y - t)

lz rn x - tln - ,u.zrl -4 r.''h-1t/i:6;9981-i

. - ' lNn- r?-lL1-9:999LIrl' _o,9981 I *o

If we use the Degnetlte-tm1l1te-lrcn trtple polnt as a startlng polnt

for lntegEtl-on, slrce thls polnt ls Hell d.etemlned themody@lca.lly

ild the conposltlon of the plrrhotlte ls stolchl.onetrlc, then N-

lE 1.0 and we get the fo1lo{Lng rsultr

(zz) ( : .oe rer)n - ( roe to)o= -(* . lz(Y - r ) (z rn N - N + 1)

* 29,+z lz radl-1 YFb;tt8-rT

(23) ( ros fo)n= -e+J?e!vv - t ) (ztn N.- N + t )

+ zs.azG ruh-1f-1-o:9t6rr -T,'831-" - 0.0435]

+ (1og fO2Jo

Tbe orygen f'ugaclty of the startlng point is controlled by the

lerctlon

(24) 3f2 Fe + 02 = r./z Fe3o4

rlth the orygen fugaclty given by

^ 4 G(25) toe to2 = fr3 nr I ac = -13r,699 + 36.93.r(oK)

for AC ln calorles (SarLon ed Sklnner, 1979), SubstitutlDgAc lnto

(25) re eet

(26) (ros ror)" = -n'flJ'+ 8.069 .

Sub6iltutlng thls result into (23), we get

(z?) roe roz= -&.y(P - r)(z rn N - N + 1) + 29.42(2 ra,r'-1y;5;96i5

_ Y1-o.99ET N\ _ e ceo _ 28,???- --i:g6l-/ f v./u> - --

Conbtnlng the 1/T tens to s1npllfy the expression, re obtain the

followlng equatlon for oxygen fugaclty as a functlon of N along the

cufle r

(zB) ros ro2 = l-za,zzz - 4+.lzQ1h N - N * r))(f - r)

+ 2e.42(2 r"or-1vEb;tt6rr *3lBF, - zr.ee

Equatlons (16) and (28) th€n 8p€clfy both Bulfur and oryg€n fugaclty

a! functlons of N, Subtltutln8 valuss of N al'loE lolnt3 along th.

oufl! to b6 cal,ou1atad,, By chooslng sufflcl€nt nunbcn of N vaLuaa,

th! curv. nly b. d.tlmlnc4 to ew Fqulr.d Prrelllon. Tablc A-1

11!t! .xdn!l.! of vaIu.! oelouleild ullng squatlonr (16) lnd (28),

Table A-1. Fugacity of oxygen and sulfur along the pyrrhotite-magnetite boundary.

5ooocLos fs" -24,89 -22.85 -2L,58 -20,53 -19.57 -78,?5 -1?.9? - t?,2! - !6.48 -r5,?8 -15,09

l o g f s , - 1 3 . L 1 - 9 . 8 0 - 8 . 1 4 - 6 . 8 2 - 5 . 6 7 - t + , & - 3 . 6 9 - 2 , & - 7 . 9 ? - 1 , 1 8 - 0 , 4 2

7o0ql.Log f6, -27,50 -L9,54 -18,J+ -17.3? -76,53 -r5,7? -15.06 -L4.J9 -L3.?5 -7J.r4 -72,55

L o g f 5 , - 1 0 . 5 4 - 8 . 0 4 - 6 . 4 9 - 5 . 2 6 - 4 , 2 ! - 3 . 2 8 - 2 , 1 , 4 - r . 6 5 - O . 9 2 - 0 , 2 3 + O , 4 28000c

ros f6, -L8,?5 -76.85 -15,71 -14,81 -14.04 -13.33 -L2.?o -72, !o - tL.5J -11.00 -10.50

L o s f s a - 9 . 1 3 - 6 , 6 1 . - 5 . I 4 - 4 , 0 0 - ) , 0 ) - 2 . 7 8 - 7 . 4 2 - o , ? 7 - 0 . 0 5 + o , y + + ! . ! 29000c

Los fs, -76,1t6 - I t+,67 -13.53 -12,68 -7L96 -LL.p, -Lo,?4 - !0,20 - 9.69 - 9.2I - 8,?.5L o s f s 2 - 7 , 8 9 - 5 , 4 2 - 4 . o 2 - 2 . 9 5 - 2 , 0 5 - 1 , 2 ? - 0 , 5 ? + 0 , 0 6 + 0 , 5 5 + 1 . 1 8 + L . 6 9

Noter Many of the above lDlnts are netastable due to the stablllty of other conlnnds such as rustlte, faJallte,pyrlte, or sulfur liquld.

WHITNEY: SULFUROUS GASES IN

Tta alf.ot of Dr.a.un o,l th. ilrr*rotlt.-m6nct1t. boundrry nay

b..! b. .rtlnrt.d by odoulrtlng th. lhlft of th. trlpl. Dotnt, lFn-

n$natlta-tst1tt.r fron xhloh th. ouna odtlnrt.!. Tha cffcot! of

plllrun on th. slolE of th. oum !n d.Irnd.nt on thr ohan6. ln

comlDsltlon of thr Brrhotltr for a gl%n su1tur tugBclty. thi!

valletlon ls not knorn, tut would cauae e ohar:ge ln the bound,arv whlch

ls s or{rr of nagnltud€ less than that causral by thr shlft Ln th€

orlg{n anal cu therfos b€ lgnond.

The suLfur fl€aclty of the trlple Flnt ls deflned. ty the

nactlon

(zil 2Fe + s2 = 2 Fes

(:o) (ros rs2)o= - rog r=fbJ:f

wheEAG ls deflneil at 1 bar IreEsuF. To nodeL a charge Ln presflr,

a PIV ten osJ be adileil to the AG of the Factlon,

z - - , P a c a v l ( P - l )(rrl \ros ,s/o =

V3d'ffi, -2A3

wr

Yh€F Avl ts the chante ln volue of the sollda ln Fetlon (29).

Thc ghlft ln lultur firgaclty ls ih.nfor! 6quatlon (3t) nfnut (30)r

Gd y'(rosrs.;P-#. ^h

rh.ra th. lalt rDproxlnatlon l! rDgroDrtrt! forl|s6. I (1OO bn or ro).

fh. orygrn fugrplty of th. trlpl. Dolnt 1r dlo off..t, *trlch

f,1Il glv. ur tlDrllnt ohrng. ln .ultu! ltg|,otty fo! |, rDoottto owt n

vdu.. Ihl! .htft td(.! th. fom

3il 9/2 N. + oz - r/z ttfu

(:rr) (rosrq.).. - losr -z7#:i-ff

. .D av ' (P- l ) _ av2P$fi A(ros rerl'- Tfu I t'i'ttror our FirDor!!, lt l! u!.firl to nprc!.nt both ohang.a r! r lhlft

ln log fg2 at e flxld. log f6a. Ilre shtft ln (tog fgr)o nu!t, th.r.-

fos b€ offs€t by a chang€ of log fS2 8long the cuffi, whlch noar th€

Btarttn€ potnt {111 b€ that for EtoLchlonstrlc FeS!

<1(1or fe^)be) Al.roe trr) 'ffi,a(rog rer)P - o

bil ao"s rsz) = - +h71o6 ts.)P

Addlng the effrcts of equatlons (lZ) aa AZ) re obtaln the followtngt

(3e) a(rog rsz)lot"r = a(roe 4)l - +fi a6o6 to"1l

- av.P av^P(3s) A(roe rse)iot"r = TF 'ffi

_ 4h ZfrE

Slrce the conpEsslbllltles ild themal exlHslons & nlnor

co@tion for sllda at cnstal lressuFs, they nay be lgnoreal sd

th€ volunes at 1 bu ura 25 oC

substltuted wLthout 61gn1f1cdrt

emr (Barton, 19?0) . The vaLues of both tems, ad the Fsuttlng

4(fog fgr)lor"f calculateil ln thls wey are tabulateal ln TabLe A-2

as a f\nctlon of tenlErature for a 1000 bar chdEg ln 1nessure, In

the equatlore of thts IEIET, the A(1og fSZ)lot"f pr kllobar rLLl be

glven the synbol ( l ) ,

PYRRHOTITE SILICIC MAGMAS

Table A-2. Pressure efect on the pyrrhotite-magnetiteboundary

Av1xlooo Av2xlooo a (l"s fsz)total

Z.rIT6 Z:F5TT ler kllobar (A)

60700&0900

1000

o. t ) )o .L790.1080.0990.091

0.0?0o,0620.o5?0,o52o,04?

0,0400,0360.0320,0300.028

Note I volwe alata frch Robte et a.I . ( 1979) ,

fhe effect of soLld solutlon of othe! catLona ln plde of lron

In !ryrhotlte ild na€netlte cu alEo be nod.elled ty conslderlng the

offset of the orlgln of the Eurye, @ely the 1rcn-nagnetlte-trcl]lte

trlple polnt. In the case of Wr*rotlie, the effects of trcn defl-

clencleE re a.lnady coEldeEd. ln the lntegatlon, eo ihe aloniHt

offect to evaLuate ls the sbstltutlon of len, The effect of

Wrhotlte solld. solutlon on aulihr fuAetty thus beconeEr

( a o ) ( r o g $ Z ) o - - l o g K + 2 1 o g . r " s

(41) l(ros fsr). = 2 1o6 5",

To a good aplEoxlnatl.on, th. .F.S nal be Epleed by the nole frutlon

of lron 1n ihe lrcn slte

N-(uz) +-" = #L = xff

"cat lons

Theefore the effect of substltutlon for lrcn on sl,fur fugelty 1s

(43) A(1og fgr). = z ros xff ,

and thoae for SOa ud H2S b€con€

(44) A(rog rsoa) = .e(ros rHas) = rog xf;o

Sollal sofutlon ln nagnetlte {111 cause a change in the orrygen

fugelty of the irlple lDlnt, dd thus the nagnetlte-pyrhotl.te curye.

Fnh equatlons (3i sa (34), tnts Flatlonshtp becones

(r*S) (los 16r)o = - log l( + r/z tot aTe*,

(4d 6(tos fer)o - t/z tos ar- ^ = r/z 1og xff^

rnere XffiOU ls the nole fmtlon of Fe3O4 in na€netlte aE deftned by

an aplropi-ate solutlon nod€l, for eHpLe stome! (1983), For our

IDrIDaes, lt ls convgnLent to represent thts 6hlft 1n orvgen fttgaclty

d lt€ equlvalent offGet of,Eulfur fugelty along the nagnetlte_

Wrhotlte boundary, a6 ru done ln the preasure coretlon (fr ana y),r / 1 ^ - 4 - ^ )

Q+il A(rog rsr) . ffi.o(1og 16.) = 6

(tre) A(los rs.)- -)/JAG'; hz) = -z/ir", ,il3*ou

The equlvalent shlfts Ln 1og fgg, ad Log fH2S @

(49) 4(rog fso.) = A(ros h"i - -t/l rog xfrae^

In a mlogous nun€r, the change 1n IDsltlon ol ,n" ,.y*rr" -

Fyrhotlte c|lru cu be calculated for use in conlFsltlons flhlch

conialn faya.Llte lather ihan na€n€tlte. ADlogous cures cu a-1.6o

be calcul-ated for oiher lron-slllcaie-Wrhotlte assembLages as loltg

as aJ.l phases cs be ch@terlzed theno(y@1cally.